Two-Step Mechanism of Membrane Disruption by Ab through Membrane
Fragmentation and Pore Formation
Michele F. M. Sciacca,†‡Samuel A. Kotler,†‡Jeffrey R. Brender,†‡Jennifer Chen,†‡Dong-kuk Lee,†§
and Ayyalusamy Ramamoorthy†‡*
†Biophysics and‡Department of Chemistry, University of Michigan, Ann Arbor, Michigan; and§Department of Fine Chemistry, Seoul National
University of Science and Technology, Seoul, Korea
mechanism by which this occurs is not fully understood. Here, we demonstrate that membrane disruption by Ab occurs by
a two-step process, with the initial formation of ion-selective pores followed by nonspecific fragmentation of the lipid membrane
during amyloid fiber formation. Immediately after the addition of freshly dissolved Ab1–40, defects form on the membrane that
share many of the properties of Ab channels originally reported from single-channel electrical recording, such as cation selec-
tivity and the ability to be blockaded by zinc. By contrast, subsequent amyloid fiber formation on the surface of the membrane
fragments the membrane in a way that is not cation selective and cannot be stopped by zinc ions. Moreover, we observed that
the presence of ganglioside enhances both the initial pore formation and the fiber-dependent membrane fragmentation process.
Whereas pore formation by freshly dissolved Ab1–40is weakly observed in the absence of gangliosides, fiber-dependent
membrane fragmentation can only be observed in their presence. These results provide insights into the toxicity of Ab and
may aid in the design of specific compounds to alleviate the neurodegeneration of Alzheimer’s disease.
Disruption of cell membranes by Ab is believed to be one of the key components of Ab toxicity. However, the
Alzheimer’s disease is a devastating neurodegenerative
disease characterized by memory loss and severe cognitive
impairment. A key pathological marker of Alzheimer’s
disease is the formation of extracellular plaques caused by
the misfolding and aggregation of Ab1-40 and Ab1-42
peptides into extended fibrillar structures known as amyloid
(1). Amyloid formation by Ab1-40is believed to be a key
early stage in the development of Alzheimer’s, as Ab1-40
aggregation has been repeatedly linked to neuronal dysfunc-
tion and death (2). The aggregation of Ab1-40generates
a complex, multifactorial response in neurons, leaving the
actual source of Ab1-40cytotoxicity unresolved. A number
of studies have identified several factors that may contribute
to the pathology of Alzheimer’s disease, including the
generation of reactive oxygen species during aggregation,
excessive aggregation of misfolded proteins leading to
stress in the endoplasmic reticulum, inflammation, disrup-
tion of cellular membrane integrity, and activation of cell
surface receptors that lead to cell death (2).
Although the exact mechanism by which Ab causes
neuronal death is not certain, one of the most clear and
consistent pathologies in Alzheimer’s disease is an elevation
of cytoplasmic Ca2þin the vicinity of Ab amyloid deposits
(3,4). The mechanism by which Ab stimulates Ca2þinflux
has not been fully elucidated, but it has been suggested
that Ab can directly disrupt membranes through the forma-
tion of ion channels (the channel hypothesis of Alzheimer’s
disease) (5). According to the channel hypothesis, small,
annular oligomers of Ab possessing a hydrophobic exterior
and hydrophilic interior insert into the membrane, spanning
the bilayer (6,7). The hollow structure of the oligomers
allows ions to cross through the hydrophilic interior of the
pore, causing an unregulated influx of Ca2þinto the cell.
Because both neurons and mitochondria are highly sensitive
to perturbations in ionic strength, a small perturbation in
intracellular calcium levels caused by the unregulated Ab
channels can trigger an apoptotic cascade (8).
Experimental support for the channel hypothesis rests
largely on atomic force microscopy (AFM), electron
microscopy, and single-channel conductance measure-
ments. Annular structures suggestive of ion channels were
directly observed by AFM when Ab1-40and Ab1-42were
reconstituted into planar lipid bilayers (7). The formation
of these annular structures correlates with single-channel
ion conductance measurements that showed stepwise
current fluctuations suggestive of the formation of discrete
pores (7,9–11). Like endogenous calcium channels, the
channels formed by Ab are charge and size selective, and
can be blockaded by specific molecules that bind to the inte-
rior of the pore, such as Znþ2(12–15). The ability of Ab
channels to be blockaded by Znþ2suggests the involvement
of a specific structure in membrane disruption, rather than
a generalized disruption of the physical integrity of the
Although the channel hypothesis accounts for many
facets of Ab toxicity through membrane disruption, some
facets remain unexplained. First, large (up to 30 nm in diam-
eter) spherical aggregates of Ab have repeatedly been found
to be toxic (16–18). This finding is surprising in light of the
channel hypothesis, which would predict that such large
Submitted May 14, 2012, and accepted for publication June 18, 2012.
Editor: Elizabeth Rhoades.
? 2012 by the Biophysical Society
702Biophysical JournalVolume 103August 2012702–710
spherical aggregates would not be able to form ion channels
unless they disassemble into smaller aggregates. Second,
aggregation of Ab is often accompanied by large-scale
morphological changes in the membrane that would not
be expected by the insertion of small oligomeric pores
(19–21). Rather, these large-scale morphological changes
suggest that physical disruption of the membrane also takes
place after prolonged incubation with Ab, which may be
a contributing factor in the toxicity of Ab. It has been
suggested that the elongation of existing fibrils adsorbed
to the membrane can remove lipids from the membrane
via a detergent-like mechanism, causing membrane disrup-
tion that could lead to cell death (22,23). However, the
relationship between these two types of membrane disrup-
tion by Ab and other amyloidogenic proteins has not been
We show here through the use of model membrane
systems that the mechanism of Ab membrane disruption
in vitro is likely to be mediated by a two-step process: 1),
before fibril formation, soluble oligomers bind to the
elongation causes membrane fragmentation through a deter-
gent-like mechanism. Furthermore, we show that the fiber-
dependent step of membrane disruption occurs only in the
presence of gangliosides, whereas pore formation occurs at
a low level in the absence of gangliosides.
MATERIALS AND METHODS
To break up any preformed aggregates, Ab1-40was initially dissolved in
NH32%v/vat a concentration of 1 mg/ml. The peptide was then lyophilized
overnight and the powder obtained was dissolved in dimethyl sulfoxide to
a final concentration of 500 mM. Each stock solution of Ab1-40was used
immediately after preparation.
Dye leakage assay
Briefly, we added 1 or 2 ml of the peptide stock solution to 100 ml
6-carboxyfluorescein-filled large unilamellar vesicles (LUVs) composed
2-oleoyl-sn-glycero-3-phosphoserine (POPS) 7:3, POPC/POPS/ganglio-
sides 5.5:3:1.5, and total lipid brain extract (TLBE) for a final peptide
concentration of 5 or 10 mM in Corning 96-well plates. Time traces were
recorded at 25?C, and the samples were shaken for 10 s before each read.
Full details of the liposome preparation and dye leakage assay can be found
in the Supporting Material.
Thioflavin T assay
The kinetics of amyloid peptide formation were measured through
increased fluorescence emission upon binding of amyloid fibers to the
commonly used amyloid-specific dye thioflavin T (ThT). Samples were
prepared by adding 1 or 2 ml of the peptide stock solution to 100 ml of
the 10 mM phosphate buffer solution (100 mM NaCl, pH 7.4, containing
10 mM ThT) containing 0.2 mg/ml LUV, to yield a final peptide concentra-
tion of 5 or 10 mM, respectively. Details regarding the liposome preparation
can be found in the Supporting Material. Experiments were carried out
simultaneously with dye leakage samples using the same microplate and
stock solution for each experiment.
We detected the presence of cation- or size-selective pores by measuring
changes in the 340:380 nm excitation ratio upon binding of Caþ2or Znþ2
to the cation-sensitive dye Fura-2 encapsulated within the LUVs. Samples
were prepared by first diluting the Fura-2 dye-filled vesicles solution with
buffer solution (10 mM Hepes buffer solution, 100 mM Fura-2 pentapotas-
sium salt, 200 mM EGTA, 100 mM NaCl, pH 7.4; to prevent the formation
of solid calcium phosphate, phosphate buffer was not used) to a final
concentration of 0.2 mg/ml. Then, Ab1-40was added with a final concentra-
tion of 5 mM or 10 mM. Fluorescencewas measured at 340 and 380 nm with
slits set for 10 nm bandwidths to obtain the baseline. After 10 min, 500 mM
of Caþ2or Znþ2were added to the sample, and changes in the 340:380 ratio
were recorded. Samples with MSI-78, an antimicrobial peptide that is
known to strongly disrupt membranes, were also run as a positive control.
Lipid sedimentation assay
We detected the presence of micelle formation by measuring the lipid
concentration in the supernatant after sedimenting 1000 nm LUVs by
centrifugation. Solutions of POPC/POPS 7:3, POPC/POPS/ganglioside
5.5:4:1.5, and TLBE LUVs at a final concentration of 1 mg/ml were
prepared as described above, except that the solutions were extruded
through 1000 nm membranes. We first incubated 20 mM Ab1-40 with
250 ml of the LUV solution for 2 days to form amyloid fibers. After incu-
bation with Ab1-40, the samples were centrifuged for 40 min at 14,000 rpm,
and the supernatant, which contained any resulting micelles, was collected,
diluted in 1 ml of chloroform, and treated as described below to measure the
lipidconcentrationin solution.SolutionswithoutAb-40wereset ascontrols.
Lipid concentrations in the supernatant were detected via the Stewart
assay, a colorimetric technique that is based on the ability of phospholipids
to form a complex with ammonium ferrothiocyanate. Initially, a calibration
curve was created for each LUV lipid composition. A series of chloroform
solutions with known concentrations of lipids (each solution 1 ml) were
prepared and treated with 1 ml of a solution containing ferric chloride
and ammonium thiocyanate. Each solution was vortexed for 20 s and the
organic phase was collected. The optical density of these standard solutions
was at 461 nm, and was plotted versus the known lipid concentration and
used as a calibration curve to determine the lipid concentrations in the
31P solid-state NMR
All of the experiments were performed on an Agilent/Varian 400 MHz
solid-state NMR spectrometer. A Varian temperature control unit was
used to maintain the sample temperature at 37?C. All31P spectra were
collected using a spin-echo sequence (90?-tau-180?-tau with tau ¼ 60 ms)
under 25 kHz two-pulse phase modulation decoupling of protons. A typical
90?pulse length of 5 ms was used with a recycle delay of 3 s. The31P chem-
ical shift spectra are referenced with respect to 85% H3PO4at 0 ppm. In
each experiment, the31P spectrum of 200 mL LUVs composed of 4 mg
POPC/POPS or POPC/POPS/gangliosidewas first acquired without peptide
in 10 mM Hepes buffer (phosphate buffer was not used due its31P signal),
100 mM NaCl, pH 7.4. After acquisition of the control spectrum without
peptide, 20 mL of 0.3 mg/ml Ab1-40or 0.1 mg/ml MSI-78 was added into
the same sample tube. Acquisition was then started after a 10 h incubation
period to obtain the spectra of the LUVs in the presence of 0.5 wt % MSI-78
or 1.6 wt % Ab1-40. Each spectrum is the result of 25,000 scans. Ab1-40was
then allowed to incubate in LUVs for 4 days before 500 mM MnCl2was
added for the paramagnetic quenching measurements.
Biophysical Journal 103(4) 702–710
Membrane Disruption by Ab
Oligomerization of Ab1-40on the membrane
strongly disrupts ganglioside-containing
We estimated the degree of membrane disruption by Ab1-40
by quantifying the dye leakage induced by Ab1-40from
vesicle-encapsulated 6-carboxyfluorescein (24). Previous
studies showed that freshly dissolved Ab1-40 typically
rupting proteins, such as antimicrobial peptides (25,26).
However, in those studies, membrane disruption was usually
measured for onlya few minutes after the addition ofAb1-40,
and therefore the time-dependent oligomerization of the
Ab1-40peptide was not taken into account (25–27).
To take into account how the ongoing aggregation of
Ab1-40may affect membrane disruption, we recorded dye
leakage from model membranes over several days after
the addition of freshly dissolved Ab1-40(Fig. 1). Corre-
sponding assays using the amyloid-sensitive dye ThT were
performed to measure fiber formation (Fig. 2). No signifi-
cant leakage was observed from membranes composed of
only POPC/POPS within 2 days, suggesting that Ab1-40
does not cause membrane defects that allow the passage
However, membrane binding by Ab1-40is highly sensitive
to membrane composition (28,29). To test a more physio-
logically relevant membrane composition, we performed
the identical experiment with lipid vesicles formed from
TLBE, which provided strikingly different results. Leakage
from TLBE vesicles was initially low, in similarity to the
leakage observed from POPC/POPS vesicles. However,
leakage from TLBE vesicles sharply increased after a lag
time of ~1000 min after the addition of Ab1-40. The time-
scale of release shares the same sigmoidal profile as amyloid
formation (Fig. 2), suggesting that this second phase of
membrane disruption is correlated with fiber formation.
However, an exact correspondence cannot be made, because
the ThTassay measures fiber formation both in solution and
on the membrane, whereas the dye leakage assay reflects
only fiber formation on the membrane, and thus the two
assays have different sensitivities (Fig. S1). To establish
correspondence between 6-carboxyfluorescein leakage and
fiber formation, we performed a seeded fiber growth assay
in the presence of TLBE LUVs (Fig. S2) (30). Neither
preformed fibers nor freshly dissolved Ab1-40caused dye
leakage; however, leakage was immediately apparent after
freshly dissolved Ab1-40was added to the preformed fibers,
establishing a direct leakage between fiber growth and
The dye release assay showed that Ab1-40disrupts TBLE
vesicles substantially more efficiently than it disrupts PC/PS
vesicles. TLBE is a mixture of a variety of phospholipids,
cholesterol, sphingolipids, and gangliosides. Gangliosides
in particular may be important for membrane disruption,
as several studies have shown that Ab binding to the
membrane surface is strongly enhanced by the ganglioside
To test whether the presence of ganglioside contributed to
membrane disruption by oligomeric Ab1-40, we performed
the same dye leakage experiment using a simpler model
membrane system (5.5:3:1.5 molar ratio of POPC/POPS/
total ganglioside extract) containing an amount of ganglio-
side similar to that found in the TLBE (33). The results for
this sample resemble those obtained from TLBE (Fig. 1,
red line), confirming that the presence of ganglioside in the
membrane is sufficient to induce membrane defects that
allow the passage of 6-carboxyfluorescein.
Cation-selective pores form immediately upon
addition of Ab to all membrane types
The results of the carboxyfluorescein assay suggest that the
level of membrane disruption after addition of Ab1-40is
initially low but becomes higher after oligomerization
begins. Although this finding is consistent with the well-
known toxicity of protofibrillar Ab, many conductance
6-carboxyfluorescein dye leakage assay. The graph illustrates the release of
6-carboxyfluorescein induced by 10 mM Ab1-40from 0.2 mg/ml POPC/
POPS LUVs 7:3 (green line), 0.2 mg/ml POPC/POPS/ganglioside LUVs
5.5/3/1.5 (red line), and 0.2 mg/ml TLBE LUVs (blue line). Dye leakage
occurs only after a lag period and is detected only in ganglioside-containing
membranes. Experiments were performed at room temperature in 10 mM
phosphate buffer, 100 mM NaCl, pH 7.4. Results are the average of three
Membrane disruption induced by Ab1-40, as measured by the
rescent emission. Thegraph shows 10 mM Ab1-40in buffer in the absence of
membranes (black line) and in the presence of 0.2 mg/ml POPC/POPS 7:3
LUVs (green line), 0.2 mg/ml POPC/POPS/ganglioside 5.5/3/1.5 LUVs
(red line), and 0.2 mg/ml TLBE LUVs (blue line). Experiments were per-
formed at room temperature in 10 mM phosphate buffer, 100 mM NaCl,
pH 7.4. Results are the average of three measurements.
Kinetics of Ab1-40amyloid formation measured by ThT fluo-
Biophysical Journal 103(4) 702–710
704Sciacca et al.
studies in planar lipid bilayers have shown channel-like
activity occurring immediately after the addition of freshly
dissolved Ab1-40(11,34–37), well before the appearance
of fibers. In addition, the pores detected by electrical
recording are cation-selective (but 6-carboxyfluorescein is
negatively charged) and are predicted on the basis of
AFM measurements to have an inner diameter of 2 nm,
which may be too small to allow efficient passage of the
6-carboxyfluorescein molecule (7). This raises the possi-
bility that Ab1-40quickly forms pores in the membrane,
but the pores are not detected by the 6-carboxyfluorescein
dye leakage assay due to its large size or the negative charge
of the molecule.
To test this possibility, we measured the influx of
ions into the LUV using vesicle-encapsulated,
calcium-sensitive Fura-2 dye. Calcium ions are smaller
than 6-carboxyfluorescein (the Stokes radius of the hydrated
Caþ2ion is 0.3 nm, compared with 0.7 nm for fluorescein)
(38,39) and positively charged, and therefore measurements
of Caþ2influx correspond more directly to conductance
measurements on planar lipid bilayers. (In principle, the
assay does not distinguish between Caþ2influx and Fura-2
efflux; however, the smaller size of Caþ2relative to Fura-
2 (molecular weight 636.5) and its positive charge make
Caþ2influx more likely.).
Fig. 3 shows the Caþ2influx obtained 30 min after incu-
bation with Ab1-40. Influx of Caþ2occurs shortly after the
addition of the peptide (within 10 min), well before
membrane disruption as measured by the 6-carboxyfluores-
cein measurement (Fig. 1), and fiber formation as measured
by ThT fluorescence (Fig. 2). The measurements in Fig. 3
were obtained by allowing Ab1-40to bind to membrane
for a short incubation period (10 min) before the addition
of Caþ2, because Caþ2may affect the affinity of Ab1-40
for the membrane (41). To better estimate the kinetics, we
performed additional experiments in TLBE LUVs by simul-
taneously adding Caþ2and Ab1-40 to the membrane
(Fig. S3). Caþ2influx was immediately detected in these
experiments; however, the total amount was lower, probably
due to the reduced affinity of Ab1-40for Caþ2containing
membranes (41). The amount of influx induced by Ab1-40
in TLBE vesicles is significant, but less than that induced
by the strongly membrane-disruptive antimicrobial peptide
MSI-78 (also known as pexiganan), which completely dis-
rupts the membrane at equivalent concentrations (Fig. 3)
(42). Interestingly, we detected Caþ2influx in POPC/POPS
LUVs, which do not contain ganglioside (Fig. 3, light gray
bars), although the amount was less than that observed
for the TLBE LUVs. This finding is in contrast to the
6-carboxyfluorescein assay, in which membrane disruption
of POPC/POPS LUVs was negligible even after prolonged
incubation. The small but measurable Caþ2influx in
peptide, but is not absolutely essential for their formation.
Zinc ions can block early permeabilization
by pores, but not fiber-dependent membrane
To confirm that Caþ2influx is correlated with the formation
of small pores similar to those detected by electrical
recording in planar bilayers, we repeated the experiment
with Fura-2 using Znþ2ions instead of Caþ2. The ability
of Zn2þto block the activity of the Ab1-40peptide channel
in single-channel conductance measurements is well estab-
lished (10,43–46) and is believed to result from the specific
binding of Znþ2to His-13 and His-14 in the interior of the
channel (47). Accordingly, Znþ2is not expected to penetrate
into LUVs if the pores detected by the Fura-2 assay are
similar to those detected by single-channel recording.
Fig. 4 shows a comparison of the results obtained with the
Fura-2 assay using Caþ2ions (red line) and Znþ2ions (blue
line). Although we clearly observed influx of Caþ2inside
the LUV immediately after the addition of Ab1-40, we did
not observe any influx of Znþ2. Because Fura-2 binds
Znþ2more tightly than it binds Caþ2(48), and MSI-78
permits the influx of both Caþ2and Znþ2(Fig. S4), this
result indicates that Znþ2, unlike Caþ2, is unable to pene-
trate into the interior LUV containing Ab1-40, even though
the two ions are similar in charge and size. This finding
suggests that the first step of membrane disruption involves
the formation of small-sized pores similar to those detected
by single-channel electrical recording.
As mentioned above, membrane disruption is initially
selective for Caþ2over carboxyfluorescein but becomes
nonselective as time progresses (compare Figs. 1 and 3).
This finding implies that two distinct mechanisms are
Ab1-40. The maximum values of Caþ2ion influx detected by encapsulated
Fura-2 after 30 min of incubation with freshly dissolved Ab1-40from
0.2 mg/ml TLBE LUVs (dark gray bars) or 0.2 mg/ml POPC/POPS 7:3
LUVs (light gray bars) are shown. In contrast to the 6-carboxyfluorescein
dye release assay, Caþ2influx occurs shortly after the addition of Ab1-40
and occurs weakly in the absence of gangliosides. The antimicrobial
peptide MSI-78 was used as a reference for total disruption of the
membrane (black bar).
Caþ2influx into LUVs after the addition of freshly dissolved
Biophysical Journal 103(4) 702–710
Membrane Disruption by Ab
operational in membrane disruption by Ab1-40. To verify
that the mechanism underlying the second step of membrane
disruption is different from the initial pore formation, and
to demonstrate that it involves fiber growth on the surface
of the bilayer, we measured the influx of ions into the
LUV caused by fiber elongation by adding freshly dissolved
Ab1-40to the solution after incubating the model membrane
with preformed fibers. As expected, the mature preformed
fibers, which do not disrupt membranes (23), did not induce
the influx of Caþ2into TLBE LUVs (Fig. 5, black line) by
themselves. However, once fiber elongation was initiated
by adding freshly dissolved Ab1-40to the preformed Ab1-40
fibers, the LUVs became permeable to both Caþ2and Znþ2.
This finding indicates that the elongation of the amyloid
fibers on the membrane surface disrupts membranes by
a mechanism different from that initially observed when
freshly dissolved Ab1-40 was added to the membrane,
because the defects in the membrane caused by fiber elonga-
tion are permeable to Znþ2.
Fiber-dependent membrane disruption occurs
by a detergent-like mechanism
The Fura-2 assay did not allow us to determine how
membrane disruption by fiber growth occurs. For other
amyloidogenic peptides, it has been demonstrated that fiber
growth is associated with the extraction of lipids from the
membrane surface, resulting in the complete fragmentation
of the membrane (49,50). Thus, it is possible that the fiber-
dependent step of membrane disruption by Ab1-40involves
a detergent-like mechanism characterized by the fragmenta-
tion of the membrane into peptide-lipid micelles or vesicles
without the appearance of defined pores.
To test this hypothesis, we incubated LUV samples with
Ab1-40for 2 days, centrifuged the samples to sediment intact
LUVs, and then performed a Stewart assay to measure lipid
concentrations in the supernatant (51). Fiber formation is
expected to be complete within the 2-day incubation time
(Fig. 2). For samples without Ab1-40 (black bars in
Fig. 6), only a small percentage of the total lipid concentra-
tion could be found in the supernatant, confirming that
almost all of the lipids had pelleted after centrifugation,
and lipids in the supernatant were likely to be the result of
membrane fragmentation by Ab1-40. Incubation with Ab1-40
caused significant membrane fragmentation of ganglioside-
containing membranes only (Fig. 6). For the POPC/POPS
The graph shows the influx of Caþ2(red line) and Zn2þ(blue line) ions
induced by Ab1-40on 0.2 mg/ml TLBE LUVs. Freshly dissolved Ab1-40
was added to each sample at time zero, and Caþ2or Znþ2was added at
600 s as indicated by the dashed line. Fura-2 is sensitive to both Caþ2
and Zn2þions, as indicated by the control membrane disruptive MSI-78
peptide (Fig. S4).
Zn2þinhibits the pore activity of freshly dissolved Ab1-40.
disruption. The figure indicates the influx of Caþ2(red line) and Zn2þ(blue
line) ions induced by adding freshly dissolved Ab1-40to 0.2 mg/ml TLBE
LUVs incubated with preformed fibers. Freshly dissolved Ab1-40was added
to each sample at time zero, and Caþ2or Znþ2was added at 600 s as indi-
cated by the dashed line. No Caþ2influx was detected after the addition of
preformed Ab1-40fibers (black line). The influx of both Caþ2and Zn2þions
(red and blue lines, respectively) were detected by seeding Ab1-40fiber
formation with preformed fiber. This finding suggests that the fiber-depen-
dent step of membrane disruption is not correlated with pore formation.
LUVs were made from TLBE lipids.
Zn2þionscannotblock the fiber-dependentstep of membrane
with Ab1-40. Lipid concentrations in the supernatant after centrifugation
from brain extract LUVs (dark gray bar) and POPC/POPS/gangliosides
5.5/3/1.5 LUVs (gray bar) after incubation with Ab1-40for 48 h are shown.
The failure of lipid vesicles to sediment is an indication of their disruption
to smaller micelle-like structures. No significant lipids were detected in the
supernatant of samples containing POPC/POPS 7:3 (light gray bar), in
agreement with the 6-carboxyfluorescein dye leakage assay (Fig. 1).
Results are the average of three independent measures, and error bars repre-
sent the standard deviation.
Membrane fragmentation induced by prolonged incubation
Biophysical Journal 103(4) 702–710
706Sciacca et al.
LUV samples (Fig. 6), the addition of Ab1-40did not elevate
the soluble fraction of lipid, in agreement with the lack of
carboxyfluorescein release from this sample (Fig. 1). This
finding suggests that membrane disruption by fiber elonga-
tion may occur by a detergent-like mechanism. Moreover,
it agrees with the results from the 6-carboxyfluorescein
dye leakage assay, which show that membrane disruption
occurs only when ganglioside is present in the membrane
The formation of small, micelle-like lipidic structures
that are characteristic of a detergent-like mechanism can
also be detected with the use of
(32,50). Fig. 7 shows the31P spectra obtained for POPC/
POPS and POPC/POPS/ganglioside LUVs before and after
the addition of Ab1-40. In each experiment, a31Pspectrum of
the LUVs was collected before and after the addition of
Ab1-40. In the absence of Ab1-40, both samples show reso-
nances around ?13 ppm and ?17 ppm originating from
PS-rich and PC-rich lipids, respectively, in the flat lamellar
phases. An isotropic peak suggesting the formation of small,
31P solid-state NMR
micelle-like lipidic structures appears after the addition of
Ab1-40only in the sample containing ganglioside (Fig. 7 B).
The corresponding POPC/POPS LUV sample does not
show an isotropic peak (Fig. 7 A), matching the results of
the sedimentation assay (Fig. 6).
Paramagnetic quenching NMR experiments
suggest that pores disappear after fiber formation
We employed paramagnetic quenching NMR experiments
to test the integrity of the membrane after fiber formation
was completed (52). Paramagnetic quenching experiments
reveal the exposure of the lipid headgroup to solvent: if
a lipid headgroup is exposed to paramagnetic Mnþ2ions,
the intensity will decrease due to paramagnetic enhanced
relaxation. If the membrane is intact, only the31P signal
from lipids in the outer leaflet will be broadened, because
the Mn2þions cannot penetrate into the LUV membrane
to quench31P signals from lipids in the inner leaflet. If pores
or other defects are present that allow Mnþ2to pass through
the lipid bilayer, both leaflets will be broadened to some
degree. Interestingly, the intensity of the main resonance
was reduced only by 50% after addition of Mnþ2to the
sample incubated with Ab1-40for 4 days (Fig. 8 B), suggest-
ing that only the headgroups of lipids in the outer leaflet of
the bilayer are exposed to Mnþ2ions after prolonged incu-
bation with Ab1-40. The resonance corresponding to the
isotropic phase was completely quenched, as expected for
a micelle-like structure. By contrast, the intensity of the
resonances corresponding to both the lamellar and isotropic
phases decreased nearly 100% after addition of the pore-
forming peptide MSI-78 (Fig. 8 A). The absence of the para-
magnetic effect of Mnþ2on lipids in the inner leaflet of
31P resonance will be broadened and
chemical shift spectra of large unilamellar vesicles composed of 7:3
POPC/POPS (A) and 5.5:3:1.5 POPC/POPS/ganglioside (B) before (blue
line) and after the addition of Ab1-40(red line) are illustrated. The small
peak near 0 ppm in the ganglioside-containing spectra indicates the forma-
tion of small, rapidly tumblinglipid structures indicativeof membrane frag-
mentation. The absence of a corresponding peak for samples without
ganglioside is an indication that membrane fragmentation does not occur.
All spectra were obtained at 37?C and referenced with respect to 85%
H3PO4at 0.0 ppm.
31P solid-state NMR of LUVs incubated with Ab1-40.31P
tected by paramagnetic quenching. (A and B)31P chemical shift spectra
of POPC/POPS/ganglioside LUVs before (top), after the addition of
MSI-78 (A) and Ab1-40(B) (middle), and after the addition of 500 mM
Mnþ2(bottom). Mnþ2completely quenches the peaks originating from
both the isotropic and lamellar phases in the MSI-78 sample, but only
partially quenches the lamellar phase in the Ab1-40sample, indicating the
absence of membrane defects after fiber formation is complete. Ab1-40
was allowed to incubate on the membrane for 4 days before acquisition.
All spectra were collected at 37?C and referenced with respect to 85%
H3PO4at 0.0 ppm.
Absence of membrane defects after fiber formation as de-
Biophysical Journal 103(4) 702–710
Membrane Disruption by Ab
LUVs containing Ab1-40is an indication of the absence of
pores or defects in the membrane after fiber formation.
Although it is widely accepted that disruption of the integ-
rity of the plasma and perhaps the mitochondrial membranes
contributes to the toxicity of amyloidogenic peptides, the
mechanism underlying this process is still not completely
understood. We demonstrate here that the membrane disrup-
tion caused by Ab1-40is a two-step process involving initial
selective disruption of the membrane by pores followed by
nonselective physical disruption during fiber formation.
Both pore formation (11) and membrane fragmentation
(20,21,32) have been observed in separate samples for
Ab1-40; however, the relationship between the two has not
The initial step of membrane disruption by Ab1-40in
LUVs shares many properties with the channels detected
by single-channel recording. First, membrane disruption is
detected in both cases very soon after the addition of freshly
dissolved Ab1-40. In LUVs, influx of Caþ2occurs immedi-
ately after the addition of Ab1-40(Fig. 4). This compares
favorably with single-channel recordings and live-cell
calcium imaging studies that showed that the influx of
Caþ2can occur several minutes after the addition of Ab1-40
(10,53). Second, both the channels observed in single-
channel recordings and the pores initially observed in
LUVs are charge selective(11). In LUVs, membrane disrup-
tion is initially characterized by the influx of positively
charged Caþ2(Figs. 3 and 4) but not the efflux of negatively
charged 6-carboxyfluorescein into the LUV (Fig. 1), despite
their relatively similar sizes. Finally, both the channel
activity in supported lipid bilayers and the initial phase of
membrane disruption in LUVs can be stopped by the addi-
tion of Znþ2(Fig. 4), most likely through the interaction of
Znþ2with His-13 and His-14 located in the inner part of
the channel (10,43–45,47). This finding suggests that pores
of a specific structure are likely to be involved in both
The second phase of membrane disruption is distin-
guished by the leakage of 6-carboxyfluorescein from the
vesicles. Initially, Ab1-40proteoliposomes of all membrane
compositions are impermeant to 6-carboxyfluorescein
(Fig. 1), and leakage can only be detected after a significant
lag time in a manner reminiscent of fiber formation (Figs. 1
and 2). A direct link between fiber formation and membrane
disruption is suggested by seeding experiments in which
membrane disruption was immediately apparent after the
addition of freshly dissolved Ab1-40to preformed fibers
Although early membrane disruption in LUVs appears to
share many properties with the channels detected by single-
channel recording, they both differ markedly from later
membrane disruption in several ways. First, later membrane
disruption is largely nonselective for the passage of mole-
cules. Although both channels detected by electrical
recording and the early phase of membrane disruption are
selective for the passage of cationic molecules, the
membrane becomes permeable to both negatively charged
6-carboxyfluorescein and Caþ2as time progresses. The
lack of selectivity in the second phase is consistent with
a total loss of the physical integrity of the membrane, result-
ing in the appearance of small, micelle-like structures as
time progresses (Figs. 6 and 7). Second, membrane damage
by fiber elongation is not stopped by the addition of Znþ2
(Fig. 5). If anything, Znþ2appears to enhance fiber-depen-
dent membrane disruption (Fig. 5). Finally, the second phase
is completely dependent on the presence of ganglioside.
Whereas early pore formation is enhanced in ganglioside-
containing membranes (Fig. 3), membrane fragmentation
by fiber elongation cannot be observed at all in the absence
of ganglioside (Figs. 6 and 7). All of these factors suggest
that a second membrane-disrupting process is not detected
in channel recording, most likely because breakage of the
membrane correlates with the formation of large, mem-
brane-bound Ab1-40oligomers (54). However, these findings
are consistent with earlier31P NMR and AFM results sug-
gesting that Ab1-40leads to the eventual disintegration of
On the basis of our data, we cannot determinewhether the
two phases of membrane disruption are truly two separate
and independent processes or a single process with several
stages. Several researchers have noted that the pores formed
by Ab1-40appear to be unstable. Instead, pores formed by
Ab1-40appear to be dynamic, with subunits of the pore
breaking off and coalescing into larger, extended, and
ThT-positive aggregates (54,55). The formation of these
larger aggregates is correlated with a large change in
conductance that is in agreement with the start of a second
(54). Furthermore, we show that membrane disruption by
Ab1-40is transient and is abolished after fiber formation is
complete (Fig. 8), suggesting the possibility that pores are
that eventually detach from the membrane. However, it is
difficult to resolve this question using ensemble techniques
that cannot follow individual oligomers on the membrane
throughout the aggregation process.
Synthetic model membranes are advantageous because
they simplify a complex system and allow hypotheses to
be tested under controlled conditions. Although the model
membranes used here are a simplification of the complex
nature of biological membranes, our data could be useful
for elucidating the mechanism underlying the toxicity of
Ab peptide to neurons. For example, it would be highly
interesting to test the synergistic effects of channel-specific
blockers such as the NA4 hexapeptide with inhibitors of
fiber elongation in inhibiting Ab (56). Experiments are
under way to test this possibility.
Biophysical Journal 103(4) 702–710
708 Sciacca et al.
Full details of liposome preparation and dye leakage assays and additional
dye leakage and fiber formation experiments are available at http://www.
This research was supported by funds from the National Institutes of Health
(GM095640 to A.R.). D.K.L. was partly supported by the Basic Science
Research Program through the National Research Foundation of Korea,
funded by the Ministry of Education, Science and Technology (2009-
1. Hardy, J. A., and G. A. Higgins. 1992. Alzheimer’s disease: the
amyloid cascade hypothesis. Science. 256:184–185.
2. Benilova, I., E. Karran, and B. De Strooper. 2012. The toxic Ab
oligomer and Alzheimer’s disease: an emperor in need of clothes.
Nat. Neurosci. 15:349–357.
3. Small, D. H. 2009. Dysregulation of calcium homeostasis in
Alzheimer’s disease. Neurochem. Res. 34:1824–1829.
4. Mattson, M. P., B. Cheng, ., R. E. Rydel. 1992. b-Amyloid peptides
destabilize calcium homeostasis and render human cortical neurons
vulnerable to excitotoxicity. J. Neurosci. 12:376–389.
5. Pollard, H. B., N. Arispe, and E. Rojas. 1995. Ion channel hypothesis
for Alzheimer amyloid peptide neurotoxicity. Cell. Mol. Neurobiol.
6. Lal, R., H. Lin,and A. P. Quist.2007.Amyloid b ionchannel: 3Dstruc-
ture and relevance to amyloid channel paradigm. Biochim. Biophys.
7. Quist, A., I. Doudevski, ., R. Lal. 2005. Amyloid ion channels:
a common structural link for protein-misfolding disease. Proc. Natl.
Acad. Sci. USA. 102:10427–10432.
8. Demuro, A., I. Parker, and G. E. Stutzmann. 2010. Calcium signaling
and amyloid toxicity in Alzheimer disease. J. Biol. Chem. 285:
9. Capone, R., H. Jang, ., R. Lal. 2012. All-D-enantiomer of b-amyloid
peptide forms ion channels in lipid bilayers. J. Chem. Theory Comput.
10. Capone, R., F. G. Quiroz, ., M. Mayer. 2009. Amyloid-b-induced ion
flux in artificial lipid bilayers and neuronal cells: resolving a contro-
versy. Neurotox. Res. 16:1–13.
11. Kagan, B. L., and J. Thundimadathil. 2010. Amyloid peptide pores and
the b sheet conformation. Adv. Exp. Med. Biol. 677:150–167.
12. Arispe, N., E. Rojas, and H. B. Pollard. 1993. Alzheimer disease
amyloid b protein forms calcium channels in bilayer membranes:
blockade by tromethamine and aluminum. Proc. Natl. Acad. Sci.
13. Diaz, J. C., O. Simakova, ., H. B. Pollard. 2009. Small molecule
blockers of the Alzheimer Ab calcium channel potently protect
neurons from Ab cytotoxicity. Proc. Natl. Acad. Sci. USA. 106:
14. Hirakura, Y., W. W. Yiu, ., B. L. Kagan. 2000. Amyloid peptide
channels: blockade by zinc and inhibition by Congo red (amyloid
channel block). Amyloid. 7:194–199.
15. Lin, H., Y. J. Zhu, and R. Lal. 1999. Amyloid b protein (1-40) forms
calcium-permeable, Zn2þ-sensitive channel in reconstituted lipid vesi-
cles. Biochemistry. 38:11189–11196.
16. Chimon, S., M. A. Shaibat, ., Y. Ishii. 2007. Evidence of fibril-like
b-sheet structures in a neurotoxic amyloid intermediate of Alzheimer’s
b-amyloid. Nat. Struct. Mol. Biol. 14:1157–1164.
17. Hoshi, M., M. Sato, ., K. Sato. 2003. Spherical aggregates of
b-amyloid (amylospheroid) show high neurotoxicity and activate tau
protein kinase I/glycogen synthase kinase-3b. Proc. Natl. Acad. Sci.
18. Solomonov, I., E. Korkotian, ., I. Sagi. 2012. Zn2þ-Ab40 complexes
form metastable quasi-spherical oligomers that are cytotoxic to
cultured hippocampal neurons. J. Biol. Chem. 287:20555–20564.
19. Morita, M., M. Vestergaard, ., M. Takagi. 2010. Real-time observa-
tion of model membrane dynamics induced by Alzheimer’s amyloid
b. Biophys. Chem. 147:81–86.
20. Michikawa, M., J. S. Gong, ., K. Yanagisawa. 2001. A novel action of
Alzheimer’s amyloid b-protein (Ab): oligomeric Ab promotes lipid
release. J. Neurosci. 21:7226–7235.
21. Yip, C. M., and J. McLaurin. 2001. Amyloid-b peptide assembly:
a critical step in fibrillogenesis and membrane disruption. Biophys. J.
22. Ikeda, K., T. Yamaguchi, ., K. Matsuzaki. 2011. Mechanism of
amyloid b-protein aggregation mediated by GM1 ganglioside clusters.
23. Williams, T. L., B. R. G. Johnson, ., L. C. Serpell. 2011. Ab42
oligomers, but not fibrils, simultaneously bind to and cause damage
to ganglioside-containing lipid membranes. Biochem. J. 439:67–77.
24. Brender, J. R., E. L. Lee, ., A. Gafni. 2011. Biphasic effects of insulin
on islet amyloid polypeptide membrane disruption. Biophys. J.
25. McLaurin, J., and A. Chakrabartty. 1996. Membrane disruption by
Alzheimer b-amyloid peptides mediated through specific binding to
either phospholipids or gangliosides. Implications for neurotoxicity.
J. Biol. Chem. 271:26482–26489.
26. Wong, P. T., J. A. Schauerte, ., A. Gafni. 2009. Amyloid-b membrane
binding and permeabilization are distinct processes influenced sepa-
rately by membrane charge and fluidity. J. Mol. Biol. 386:81–96.
27. Gauci, A. J., M. Caruana, ., N. Vassallo. 2011. Identification of poly-
phenolic compounds and black tea extract as potent inhibitors of lipid
membrane destabilization by Ab(42) aggregates. J. Alzheimers Dis.
28. Matsuzaki, K. 2007. Physicochemical interactions of amyloid
29. Sani, M. A., J. D. Gehman, and F. Separovic. 2011. Lipid matrix plays
a role in Ab fibril kinetics and morphology. FEBS Lett. 585:749–754.
30. Engel, M. F., L. Khemte ´mourian, ., J. W. Ho ¨ppener. 2008. Membrane
damage by human islet amyloid polypeptide through fibril growth at
the membrane. Proc. Natl. Acad. Sci. USA. 105:6033–6038.
31. Ikeda, K., and K. Matsuzaki. 2008. Driving force of binding of amyloid
b-protein to lipid bilayers. Biochem. Biophys. Res. Commun. 370:
32. Nakazawa, Y., Y. Suzuki, ., T. Asakura. 2009. The interaction of
amyloid Ab (1-40) with lipid bilayers and ganglioside as studied by
31P solid-state NMR. Chem. Phys. Lipids. 158:54–60.
33. Jensen, M., and W. Nerdal. 2008. Anticancer cisplatin interactions with
bilayers of total lipid extract from pig brain: A13C, 31P and 15N solid-
state NMR study. Eur. J. Pharm. Sci. 34:140–148.
34. Inoue,S. 2008. Insitu Ab poresin AD brain are cylindricalassembly of
Ab protofilaments. Amyloid. 15:223–233.
35. Jang, H., F. T. Arce, ., R. Nussinov. 2010. b-Barrel topology of
Alzheimer’s b-amyloid ion channels. J. Mol. Biol. 404:917–934.
36. Kayed, R., A. Pensalfini, ., C. Glabe. 2009. Annular protofibrils are
a structurally and functionally distinct type of amyloid oligomer.
J. Biol. Chem. 284:4230–4237.
37. Small, D. H., R. Gasperini, ., L. Foa. 2009. The role of Ab-induced
calcium dysregulation in the pathogenesis of Alzheimer’s disease.
J. Alzheimers Dis. 16:225–233.
38. Corti, H. R., G. A. Frank, and M. C. Marconi. 2008. An alternate
solution of fluorescence recovery kinetics after spot-bleaching for
measuring diffusion coefficients. 2. Diffusion of fluorescein in aqueous
sucrose solutions. J. Solution Chem. 37:1593–1608.
39. Maroudas, A., P. D. Weinberg, ., C. P. Winlove. 1988. The distribu-
tions and diffusivities of small ions in chondroitin sulphate, hyaluro-
nate and some proteoglycan solutions. Biophys. Chem. 32:257–270.
Biophysical Journal 103(4) 702–710
Membrane Disruption by Ab
40. Reference deleted in proof.
41. Meratan, A. A., A. Ghasemi, and M. Nemat-Gorgani. 2011. Membrane
integrity and amyloid cytotoxicity: a model study involving mitochon-
dria and lysozyme fibrillation products. J. Mol. Biol. 409:826–838.
42. Hallock, K. J., D. K. Lee, and A. Ramamoorthy. 2003. MSI-78, an
analogue of the magainin antimicrobial peptides, disrupts lipid bilayer
structure via positive curvature strain. Biophys. J. 84:3052–3060.
43. Capone, R., H. Jang, ., R. Lal. 2012. Probing structural features of
Alzheimer’s amyloid-b pores in bilayers using site-specific amino
acid substitutions. Biochemistry. 51:776–785.
44. Hirakura, Y., M. C. Lin, and B. L. Kagan. 1999. Alzheimer amyloid
ab1-42 channels: effects of solvent, pH, and Congo Red. J. Neurosci.
45. Arispe, N., H. B. Pollard, and E. Rojas. 1996. Zn2þinteraction with
Alzheimer amyloid b protein calcium channels. Proc. Natl. Acad.
Sci. USA. 93:1710–1715.
46. Kawahara, M., N. Arispe, ., E. Rojas. 1997. Alzheimer’s disease
amyloid b-protein forms Zn(2þ)-sensitive, cation-selective channels
across excised membrane patches from hypothalamic neurons.
Biophys. J. 73:67–75.
47. Dı ´az, J. C., J. Linnehan, ., N. Arispe. 2006. Histidines 13 and 14 in
the Ab sequence are targets for inhibition of Alzheimer’s disease Ab
ion channel and cytotoxicity. Biol. Res. 39:447–460.
48. Grynkiewicz, G., M. Poenie, and R. Y. Tsien. 1985. A new generation
of Ca2þindicators with greatly improved fluorescence properties.
J. Biol. Chem. 260:3440–3450.
49. Sparr, E., M. F. Engel, ., J. A. Killian. 2004. Islet amyloid polypep-
tide-induced membrane leakage involves uptake of lipids by forming
amyloid fibers. FEBS Lett. 577:117–120.
50. Brender, J. R., U. H. N. Du ¨rr, ., A. Ramamoorthy. 2007. Membrane
fragmentation by an amyloidogenic fragment of human islet amyloid
polypeptide detected by solid-state NMR spectroscopy of membrane
nanotubes. Biochim. Biophys. Acta. 1768:2026–2029.
51. Stewart, J. C. M. 1980. Colorimetric determination of phospholipids
with ammonium ferrothiocyanate. Anal. Biochem. 104:10–14.
52. Last, N. B., E. Rhoades, and A. D. Miranker. 2011. Islet amyloid
polypeptide demonstrates a persistent capacity to disrupt membrane
integrity. Proc. Natl. Acad. Sci. USA. 108:9460–9465.
53. Simakova, O., and N. J. Arispe. 2006. Early and late cytotoxic effects
of external application of the Alzheimer’s Ab result from the initial
formation and function of Ab ion channels. Biochemistry. 45:5907–
54. Schauerte, J. A., P. T. Wong, ., A. Gafni. 2010. Simultaneous single-
molecule fluorescence and conductivity studies reveal distinct classes
of Ab species on lipid bilayers. Biochemistry. 49:3031–3039.
55. Jang, H., F. T. Arce, ., R. Nussinov. 2009. Misfolded amyloid ion
channels present mobile b-sheet subunits in contrast to conventional
ion channels. Biophys. J. 97:3029–3037.
56. Arispe, N., J. C. Diaz, and O. Simakova. 2007. Ab ion channels.
Prospects for treating Alzheimer’s disease with Ab channel blockers.
Biochim. Biophys. Acta. 1768:1952–1965.
Biophysical Journal 103(4) 702–710
710Sciacca et al.